Combinational feature based random forest classification for enhanced bundle branch block beat detection

Automated detection and classification of a bundle branch block (BBB) beat increase the chances of early diagnosis and the possibility of better and timely treatment. However, latest BBB classifiers reach their bottleneck when the number of features increases or the size of training data is limited. In this paper, we apply a random forest classifier to overcome the shortcomings, but also propose the use of multiple distinctive features, which are extracted from the ECG patterns by principal component analysis (PCA), magnitude squared coherence (MSC), and wavelet transform, to further enhance the performance. The testing results demonstrate the effectiveness of the proposed methodology by showing an outstanding detection performance compared to other traditional methods.